The diagnosis and management of normal pressure hydrocephalus (NPH) remains controversial. The aim of this prospective study was to assess the usefulness of clinical and radiological criteria together with supplemental neuropsychological and gait tests, cerebrospinal fluid (CSF) hydrodynamic studies and external lumbar drainage (ELD) in identifying those who may respond to a shunt and to compare the outcomes with the non-shunted patients at one year. Method: Forty patients with a clinical diagnosis of NPH were prospectively studied according to a fixed management protocol. Resistance to CSF (Rcsf) was measured using a lumbar infusion study and an ELD was used to determine improvement in neuropsychological and gait tests following CSF drainage. Based on specific criteria those who showed improvement were shunted. Clinical and radiological outcomes were assessed at one year in all patients. Results: Twenty three (57.5%) patients were shunted. Improvement was observed in 74% of shunted patients, while 17% did not improve and 9% deteriorated following surgery. Age, etiology, presentation, duration of symptoms and presence of co-morbid factors were unrelated to outcome. Improvement was found in 63% of shunted patients with RCSF of 12 mmHg/ml/min or higher. The sensitivity was 64% with a positive predictive value of 68%. Both Rcsf testing and ELD enhanced the positive predictive outcome of shunt operation. Using the non-shunted patients as controls, the mean difference between the two groups over time differed significantly in all the neuropsychological tests and some gait test. Conclusions: No single test was able to predict overall success with shunting but the results showed that a high percentage of improvement could be achieved by using continuous ELD and a rigorous protocol. Greater improvements were observed in cognitive and gait function than in sphincter control. An Rcsf of 12mmHg/ml/min or more was related to better outcomes. Consideration was given to the significant limitations in the study design and outcome measures.